Optimizing Airport Traffic Flow Management
Uploaded by
raonerahul001Optimizing Airport Traffic Flow Management
Uploaded by
raonerahul001Common questions
Powered by AISpatial and temporal feature extraction mechanisms, such as Convolutional Neural Networks (CNN), Temporal Convolutional Networks (TCN), and attention mechanisms, improve ATFM delay prediction models by extracting representative features and mining hidden information from multidimensional prediction data. CNN captures spatial features like airspace distribution, while TCN captures long-term dependencies and temporal correlations to model nonlinear relationships, enhancing prediction accuracy .
The ATFM delay prediction accuracy might vary for different airports due to differences in actual delay values and the complexity of local factors affecting delays. In the study, the SSA-LSTM-1 prediction model showed better accuracy for ZSNJ (Nanjing Lukou International Airport) and ZSPD (Shanghai Pudong International Airport) compared to ZSHC (Hangzhou Xiaoshan International Airport) and ZSSS (Shanghai Hongqiao International Airport), particularly for delays over 60 minutes .
The ATFM delay coefficient (D) is calculated by comparing the calculated takeoff time (CTOT) and the true takeoff time (TTOT) of a flight. If CTOT-TTOT is less than or equal to zero, D is 0, indicating no ATFM delay; otherwise, D is 1, indicating the flight experiences an ATFM delay. This coefficient helps in identifying which flights are experiencing delays at specific stages .
The study concludes that improved deep learning algorithms can effectively address problems associated with multi-source, high-dimensional, and unbalanced ATFM delay prediction data, enabling short-term ATFM delay predictions. These algorithms construct a dynamic prediction network, taking into account capacity and demand imbalances, and provide more accurate predictions by using nodes and edges to represent airports and waypoints .
The importance of nodes such as airports or key waypoints is used in predicting ATFM delays by calculating their PageRank values in the ATFM delay prediction network. A higher PageRank value indicates greater importance and a higher probability of experiencing ATFM delay, which helps in identifying critical points in the network requiring focused delay management .
The study validates its ATFM delay prediction model by applying it to selected domestic routes in a highly congested region in East China and comparing predicted ATFM delay values to actual values. Key findings indicate that the model's predicted values are generally lower than actual values, with better accuracy for delays shorter than 60 minutes. The prediction accuracy was higher for certain airports, such as ZSNJ and ZSPD, revealing inconsistencies in performance across different congestion levels .
The Dynamic Weighted PageRank algorithm is used to calculate the importance of the nodes in an ATFM delay prediction network, providing a more accurate indication of ATFM delay probability by reflecting edge weights and temporal factors. It is preferred over the traditional PageRank algorithm because it addresses the average distribution irrationality, making it more applicable to dynamically weighted networks like those found in air traffic management .
To address inconsistencies between ATFM delay generation and occurrence locations, the study constructs a dynamic ATFM delay prediction network model where nodes represent points of capacity-demand imbalance and edges represent routes. This approach allows for identifying potential delay areas, rather than relying on isolated points, thus handling inconsistencies more effectively .
Infrastructure development alone is insufficient to address growing air traffic delays because it is very costly and takes many years to implement. Additionally, experts agree that it will not be enough to limit increases in delays above current levels. As an alternative, the thesis suggests incorporating optimization into air traffic flow management (ATFM) to minimize delays, reduce emissions, and improve safety management .
The thesis proposes unified approaches to optimizing traffic flow at airports on both strategic and tactical levels. These include selecting a runway configuration sequence, balancing arrivals and departures, assigning flights to runways and determining their sequence, gate-holding durations for departures, speed-control for arrivals, and routing flights within the terminal area. These approaches are significant as they address the entirety of key problems faced at an airport, offering a holistic solution rather than optimizing single aspects at a time .